Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Use new LASR in STEQR and BDSQR and parallelize STEQR #897

Open
wants to merge 7 commits into
base: develop
Choose a base branch
from

Conversation

tfalders
Copy link
Collaborator

No description provided.

@tfalders tfalders added the noOptimizations Disable optimized kernels for small sizes for some routines label Feb 21, 2025
@jzuniga-amd
Copy link
Collaborator

jzuniga-amd commented Feb 21, 2025

Looks good. Adding a comment below to trigger the performance tracking system (PTS) so that we can verify performance improvements.

@jzuniga-amd
Copy link
Collaborator

%PTS
gesvd

@jzuniga-amd jzuniga-amd added the runPerformance Run CI performance job label Feb 21, 2025
@tfalders tfalders changed the title Use new LASR in STEQR and BDSQR Use new LASR in STEQR and BDSQR and parallelize STEQR Mar 3, 2025
{
D[end] *= scale;
for(int i = start; i < end; i++)
for(int i = tid; i < end; i += tid_inc)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

and how is start taken into consideration now?
The old code scales elements in [start, end]
This new multithreaded code scales elements in [first_tid, end]... what if the first thread index is not equal to start?

if(anorm == 0)
continue;
else if(anorm > ssfmax)
scale_tridiag(l, lend, D, E, anorm / ssfmax);
scale_tridiag(lsv, lendsv, D, E, anorm / ssfmax, tid, tid_inc);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

these scale_tridiag calls are being done by all threads (from tid=0 to tid=63) to scale elements in [lsv, lendsv]
I don't understand what happened when lsv != 0 ?
(See the comments about scale_tridiag above)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
noOptimizations Disable optimized kernels for small sizes for some routines runPerformance Run CI performance job
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants